Creating multiple graphs using gnp_random_graph

While using the NetworkX package I was tasked with creating multiple random graphs with a given n number of nodes and p probability, this is my code:

def random_networks_generator(n,p,num_networks=1, directed=False,seed=30030390):
    Graph_list=[]
    for num in range (0,num_networks):
        G=nx.gnp_random_graph(n,p,seed,directed)
        Graph_list.append(G)
    return Graph_list

But, every iteration creates the same exact graph (even the edges are completely the same)

Does anyone have a clue what might be wrong?

Update:

After trying to use the function without the “seed” parameter the graphs are random, but is there a way to sort the problem while still using the “seed” parameter?

Answer

Currently you make all the calls in the loop with the same fixed seed. According to the documentation of gnp_random_graph or more general Randomness by fixing the seed you make sure that you always receive the exactly same random graph.

For example:

G_1=nx.gnp_random_graph(100, .5, 42)
G_2=nx.gnp_random_graph(100, .5, 42)
G_3=nx.gnp_random_graph(100, .5, 43)

The graphs G_1 and G_2 will have exactly the same edges, whereas G_3 got a different seed and with high likelihood is different (as it is in this example).

You can get the desired behaviour of retrieving the same list of graphs (determined by the initial seed) via the following code:

def random_networks_generator(n, p, num_networks=1, directed=False, seed=30030390):
    Graph_list=[]
    for num in range (0,num_networks):
        G=nx.gnp_random_graph(n, p, seed + num, directed)
        Graph_list.append(G)
    return Graph_list

More background on seed:

• What is the use of numpy.random.seed() Does it make any difference?
• random.seed(): What does it do?